Skip to contentSkip to conversation

csky: Fix a size determination in gpr_get()

PATCH In review Zhenzhong Duan 3 participants 11 messages Sep 22, 2020 – Dec 24, 2020

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 csky: Fix a size determination in gpr_get()

"*" is missed in size determination as we are passing register set rather than a pointer.

Fixes dcad7854fcce ("sky: switch to ->regset_get()")
---
 arch/csky/kernel/ptrace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/csky/kernel/ptrace.c +1 -1
index d822144906ac..a4cf2e2ac15a 100644
@@ -83,7 +83,7 @@ static int gpr_get(struct task_struct *target,
 	/* Abiv1 regs->tls is fake and we need sync here. */
 	regs->tls = task_thread_info(target)->tp_value;
 
-	return membuf_write(&to, regs, sizeof(regs));
+	return membuf_write(&to, regs, sizeof(*regs));
 }
 
 static int gpr_set(struct task_struct *target,
Signature · 1 line
2.25.1
#1 lore raw
Quoted 2 lines · Zhenzhong Duan

"*" is missed in size determination as we are passing register set rather than a pointer.

Ack. I can push it to Linus today, unless you want it to go through csky tree. Preferences?

Quoted 22 lines
Fixes dcad7854fcce ("sky: switch to ->regset_get()")
---
 arch/csky/kernel/ptrace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/csky/kernel/ptrace.c +1 -1
index d822144906ac..a4cf2e2ac15a 100644
@@ -83,7 +83,7 @@ static int gpr_get(struct task_struct *target,
 	/* Abiv1 regs->tls is fake and we need sync here. */
 	regs->tls = task_thread_info(target)->tp_value;
 
-	return membuf_write(&to, regs, sizeof(regs));
+	return membuf_write(&to, regs, sizeof(*regs));
 }
 
 static int gpr_set(struct task_struct *target,
Signature · 1 line
2.25.1
maintainer Acked-by
#2 lore raw

Thx Duan,

Hi AI,

I found the broken commit still has a question:

Quoted 7 lines

commit dcad7854fcce6a2d49b6a3ead5bbefeff047e559

Author: Al Viro <[email protected]>
Date:   Tue Jun 16 15:28:29 2020 -0400

csky: switch to ->regset_get()

NB: WTF "- what the fuck :(" is fpregs_get() playing at???

The fpregs_get() is for REGSET_FPR regset used by ptrace (gdb) and all fp regs are stored in threads' context.
So, WTF question for?

Best Regards
Guo Ren

Quoted 30 lines · Al Viro
Quoted 2 lines · Zhenzhong Duan

"*" is missed in size determination as we are passing register set rather than a pointer.

Ack. I can push it to Linus today, unless you want it to go through csky tree. Preferences?

Quoted 22 lines
Fixes dcad7854fcce ("sky: switch to ->regset_get()")
---
 arch/csky/kernel/ptrace.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/csky/kernel/ptrace.c +1 -2
index d822144906ac..a4cf2e2ac15a 100644
@@ -83,7 +83,7 @@ static int gpr_get(struct task_struct *target,
      /* Abiv1 regs->tls is fake and we need sync here. */
      regs->tls = task_thread_info(target)->tp_value;
 
-     return membuf_write(&to, regs, sizeof(regs));
+     return membuf_write(&to, regs, sizeof(*regs));
 }
 
 static int gpr_set(struct task_struct *target,
--

2.25.1

Signature · 4 lines
Best Regards
 Guo Ren

ML: https://lore.kernel.org/linux-csky/
#3 lore raw
Quoted 12 lines · Guo Ren

Thx Duan,

Hi AI,

I found the broken commit still has a question:

Quoted 3 lines

commit dcad7854fcce6a2d49b6a3ead5bbefeff047e559

Author: Al Viro <[email protected]>
Date:   Tue Jun 16 15:28:29 2020 -0400
Quoted 6 lines
Quoted 3 lines

csky: switch to ->regset_get()

NB: WTF "- what the fuck :(" is fpregs_get() playing at???

The fpregs_get() is for REGSET_FPR regset used by ptrace (gdb) and all fp regs are stored in threads' context.
So, WTF question for?

The part under
#if defined(CONFIG_CPU_HAS_FPUV2) && !defined(CONFIG_CPU_HAS_VDSP)

What's going on there? The mapping is really weird - assuming you had v0..v31 in the first 32 elements of regs->vr[], you end up with

v0 v1 v2 v3 v2 v3 v6 v7 v4 v5 v10 v11 v6 v7 v14 v15 v8 v9 v18 v19 v10 v11 v22 v23 v12 v13 v26 v27 v14 v15 v30 v31

in the beginning of the output. Assuming it is the intended behaviour, it's probably worth some comments...

maintainer
#4 lore raw
Quoted 28 lines · Al Viro
Quoted 12 lines · Guo Ren

Thx Duan,

Hi AI,

I found the broken commit still has a question:

Quoted 3 lines

commit dcad7854fcce6a2d49b6a3ead5bbefeff047e559

Author: Al Viro <[email protected]>
Date:   Tue Jun 16 15:28:29 2020 -0400
Quoted 6 lines
Quoted 3 lines

csky: switch to ->regset_get()

NB: WTF "- what the fuck :(" is fpregs_get() playing at???

The fpregs_get() is for REGSET_FPR regset used by ptrace (gdb) and all fp regs are stored in threads' context.
So, WTF question for?

The part under
#if defined(CONFIG_CPU_HAS_FPUV2) && !defined(CONFIG_CPU_HAS_VDSP)

What's going on there? The mapping is really weird - assuming you had v0..v31 in the first 32 elements of regs->vr[], you end up with

Quoted 5 lines

v0 v1 v2 v3 v2 v3 v6 v7 v4 v5 v10 v11 v6 v7 v14 v15 v8 v9 v18 v19 v10 v11 v22 v23 v12 v13 v26 v27 v14 v15 v30 v31

in the beginning of the output. Assuming it is the intended behaviour, it's probably worth some comments...

FPU & VDSP use the same regs. 32 FPU regs' width is 64b and 16 VDSP regs' width is 128b.

vr[0], vr[1] = fp[0] & vr[0] vr[1], vr[2], vr[3] = vdsp reg[0]
...
vr[60], vr[61] = fp[15] & vr[60] vr[61], vr[62], vr[63] = vdsp reg[15] vr[64], vr[65] = fp[16]
vr[66], vr[67] = fp[17]
...
vr[94], vr[95] = fp[31]

Yeah, this is confusing and I'll add a comment later.

--
Best Regards
Guo Ren

ML: https://lore.kernel.org/linux-csky/

Acked-by
#5 lore raw
Quoted 20 lines · Guo Ren
Quoted 9 lines

What's going on there? The mapping is really weird - assuming you had v0..v31 in the first 32 elements of regs->vr[], you end up with

v0 v1 v2 v3 v2 v3 v6 v7 v4 v5 v10 v11 v6 v7 v14 v15 v8 v9 v18 v19 v10 v11 v22 v23 v12 v13 v26 v27 v14 v15 v30 v31

in the beginning of the output. Assuming it is the intended behaviour, it's probably worth some comments...

FPU & VDSP use the same regs. 32 FPU regs' width is 64b and 16 VDSP regs' width is 128b.

vr[0], vr[1] = fp[0] & vr[0] vr[1], vr[2], vr[3] = vdsp reg[0]
...
vr[60], vr[61] = fp[15] & vr[60] vr[61], vr[62], vr[63] = vdsp reg[15] vr[64], vr[65] = fp[16]
vr[66], vr[67] = fp[17]
...
vr[94], vr[95] = fp[31]

Quoted 1 line

Yeah, this is confusing and I'll add a comment later.

Umm... It would help if you described these 3 layouts:
1) kernel-side with VDSP
2) userland (identical to (1)?)
3) kernel-side without VDSP
Still confused...

PS: my apologies re commit message - I left a note to myself when doing that series and then forgot about it ;-/

Anyway, which tree should it go through? In any case, that fix is

and I can take it through vfs.git or you guys can pick in csky tree; up to you.

maintainer
#6 lore raw
Quoted 23 lines · Al Viro
Quoted 20 lines · Guo Ren
Quoted 9 lines

What's going on there? The mapping is really weird - assuming you had v0..v31 in the first 32 elements of regs->vr[], you end up with

v0 v1 v2 v3 v2 v3 v6 v7 v4 v5 v10 v11 v6 v7 v14 v15 v8 v9 v18 v19 v10 v11 v22 v23 v12 v13 v26 v27 v14 v15 v30 v31

in the beginning of the output. Assuming it is the intended behaviour, it's probably worth some comments...

FPU & VDSP use the same regs. 32 FPU regs' width is 64b and 16 VDSP regs' width is 128b.

vr[0], vr[1] = fp[0] & vr[0] vr[1], vr[2], vr[3] = vdsp reg[0]
...
vr[60], vr[61] = fp[15] & vr[60] vr[61], vr[62], vr[63] = vdsp reg[15] vr[64], vr[65] = fp[16]
vr[66], vr[67] = fp[17]
...
vr[94], vr[95] = fp[31]

Quoted 4 lines
Quoted 1 line

Yeah, this is confusing and I'll add a comment later.

Umm... It would help if you described these 3 layouts:
1) kernel-side with VDSP

With VDSP: we use vdsp ld/st instructions to access first 16 128bit-regs and use fpu ld/st instructions to access last 16 64bit-regs.

Quoted 1 line

2) userland (identical to (1)?)

Identical to 1.

Quoted 1 line

3) kernel-side without VDSP

Without VDSP: we use fpu ld/st instructions to access last 32 64bit-regs.

So, there are 96 32bit-vr[] for the struct. And with VDSP or not will got different storage format.
With VDSP:
vr128[16] // contain first fp64[16]
fp64[16]; // second fp64[16]

Without VDSP:
fp64[32]
no-use for the reset vr[]

Quoted 6 lines

Still confused...

PS: my apologies re commit message - I left a note to myself when doing that series and then forgot about it ;-/

Anyway, which tree should it go through? In any case, that fix is

Thx for your job, and pushing to linus with Zhenzhong Duan's advice.

Quoted 3 lines

and I can take it through vfs.git or you guys can pick in csky tree; up to you.

--
Best Regards
Guo Ren

ML: https://lore.kernel.org/linux-csky/

#7 lore raw
In reply to Al Viro
Quoted 31 lines · Al Viro
Quoted 20 lines · Guo Ren
Quoted 9 lines

What's going on there? The mapping is really weird - assuming you had v0..v31 in the first 32 elements of regs->vr[], you end up with

v0 v1 v2 v3 v2 v3 v6 v7 v4 v5 v10 v11 v6 v7 v14 v15 v8 v9 v18 v19 v10 v11 v22 v23 v12 v13 v26 v27 v14 v15 v30 v31

in the beginning of the output. Assuming it is the intended behaviour, it's probably worth some comments...

FPU & VDSP use the same regs. 32 FPU regs' width is 64b and 16 VDSP regs' width is 128b.

vr[0], vr[1] = fp[0] & vr[0] vr[1], vr[2], vr[3] = vdsp reg[0]
...
vr[60], vr[61] = fp[15] & vr[60] vr[61], vr[62], vr[63] = vdsp reg[15] vr[64], vr[65] = fp[16]
vr[66], vr[67] = fp[17]
...
vr[94], vr[95] = fp[31]

Quoted 1 line

Yeah, this is confusing and I'll add a comment later.

Umm... It would help if you described these 3 layouts:
1) kernel-side with VDSP
2) userland (identical to (1)?)
3) kernel-side without VDSP
Still confused...

Quoted 7 lines

PS: my apologies re commit message - I left a note to myself when doing that series and then forgot about it ;-/

Anyway, which tree should it go through? In any case, that fix is

and I can take it through vfs.git or you guys can pick in csky tree; up to you.

Hi Al, Guo

Seems this patch is still pending, could you help check it?Thanks

Zhenzhong

maintainer
#8 lore raw

Hi Zhengzhong,

I'll take it, thx.

Quoted 46 lines · Zhenzhong Duan
Quoted 31 lines · Al Viro
Quoted 20 lines · Guo Ren
Quoted 9 lines

What's going on there? The mapping is really weird - assuming you had v0..v31 in the first 32 elements of regs->vr[], you end up with

v0 v1 v2 v3 v2 v3 v6 v7 v4 v5 v10 v11 v6 v7 v14 v15 v8 v9 v18 v19 v10 v11 v22 v23 v12 v13 v26 v27 v14 v15 v30 v31

in the beginning of the output. Assuming it is the intended behaviour, it's probably worth some comments...

FPU & VDSP use the same regs. 32 FPU regs' width is 64b and 16 VDSP regs' width is 128b.

vr[0], vr[1] = fp[0] & vr[0] vr[1], vr[2], vr[3] = vdsp reg[0]
...
vr[60], vr[61] = fp[15] & vr[60] vr[61], vr[62], vr[63] = vdsp reg[15] vr[64], vr[65] = fp[16]
vr[66], vr[67] = fp[17]
...
vr[94], vr[95] = fp[31]

Quoted 1 line

Yeah, this is confusing and I'll add a comment later.

Umm... It would help if you described these 3 layouts:
1) kernel-side with VDSP
2) userland (identical to (1)?)
3) kernel-side without VDSP
Still confused...

Quoted 7 lines

PS: my apologies re commit message - I left a note to myself when doing that series and then forgot about it ;-/

Anyway, which tree should it go through? In any case, that fix is

and I can take it through vfs.git or you guys can pick in csky tree; up to you.

Hi Al, Guo

Seems this patch is still pending, could you help check it?Thanks

Zhenzhong

Signature · 4 lines
Best Regards
 Guo Ren

ML: https://lore.kernel.org/linux-csky/
#9 lore raw
Quoted 4 lines · Guo Ren

Hi Zhengzhong,

I'll take it, thx.

Thanks Guo.

Zhenzhong

#10 lore raw
In reply to Al Viro
Quoted 7 lines · Al Viro
Quoted 2 lines · Zhenzhong Duan

"*" is missed in size determination as we are passing register set rather than a pointer.

Ack. I can push it to Linus today, unless you want it to go through csky tree. Preferences?

I prefer pushing to linus.

Regards
Zhenzhong